Name: Piyush Gupta
Type: User
Company: University of Minnesota
Bio: Data Scientist - Analytics professional with more than ~4 years of experience working with multiple Fortune 500 firms across CPG/, Banking, financial services
Location: Minneapolis
Blog: https://guptapiyush340.github.io/website/
Piyush Gupta's Projects
Automating Data Science pipeline using airflow
Basic EDA & data manipulation for Big Mart Sales
Building model using GBM and XGBoost in Python
Visualizing how predictive algorithms works can be challenging at times. With more focus on mathematics, we tend to miss out of intuition behind these models. Generating decision boundaries for different hyper parameter values can not only improve our understanding but also help to overcome over fitting..
Building different model and creating ensemble of models
Implementation of boosting algorithms in PySpark with local Spark cluster
Implementation of boosting algorithms in SparklyR with local Spark cluster
Building neural network and performing image processing with the help of MNIST dataset
Learning k-means clustering and it performance improvement
Practiced CNN implementation using Kaggle competition of Dogs vs Cats
Using LightGBM and NLP to predict the application approval for Kaggle Competitions DonorsChoose.org
Directory of tutorials and open-source code repositories for working with Keras, the Python deep learning library
Building multi nomial logistic regression model using glass dataset
Comparison among models can be tricky and tedious task due to multiple hyper parameter settings and random split of data. Nested cross validation resolves the issue by selecting best performing algorithm overall.
Python code for Causal inference mix tape by Prof. Scott Cunningham. https://mixtape.scunning.com/potential-outcomes.html
This project includes following repositories Presentation Machine Learning algorithms like Prophet, ARIMA, XGBoost, LSTM and Seq2Seq
Generating Tabular Synthetic Data using State of the Art GAN architecture
Generative adversarial training for synthesizing tabular data
This repo aims to be a useful collection of notebooks/code for understanding and implementing seq2seq neural networks for time series forecasting. Networks are constructed with keras/tensorflow.
https://guptapiyush340.github.io/website/